基于时变特征和核非线性分类器的飞机目标识别

Aircraft Recognition Based on Time-Frequency Features and Kernel-Based Nonlinear Classifier

基于雷达目标距离像,研究时变特征提取和核分类器在雷达目标识别中的应用。由于距离像敏感于目标姿态角的变化,单纯的时域或频域方法难以完整刻画目标的散射特性,因此文中采用时频分析方法,首先提取出距离像时频分布的特征参量,再利用主元分析法降低维数,最后采用基于核的非线性分类器进行目标识别。仿真数据和实测数据表明,该方法具有较好的识别效果。

Time-frequency analysis and kernel-based nonlinear classifiers are applied to aircraft recognition using high resolution range profiles（HRRP）. Since HRRP is sensitive to the variation of target aspect, methods discussed only in time or frequency domain cannot completely represent the scattering characteristics of target. For this reason, we propose in this paper to extract features from time-frequency distribution of a profile and to reduce the dimension by principal component analysis（PCA）. And then, kernel-based nonlinear classifiers are applied to classify the features. Experimental results on both simulated and measured profiles show comparatively good performance of the method.